S. G. Emelyanov - Dr.Sc. (Eng.), Professor, Rector, South-West State University (Kursk)
S. N. Frolov - Post-graduate Student, Department "Information systems und Technologies", South-West State University (Kursk)
S. Yu. Sazonov - Ph.D. (Eng.), Associate Professor, Department "Information systems und Technologies", South-West State University (Kursk)

Recently in systems of electroensuring technically difficult objects there was rather unsuccessful state concerning their fire-prevention stability, owing to an unsatisfactory condition of fixed assets, slow rates of reconstruction, lack of reliable systems of the prevention and localization of fire-dangerous situations, control and protection devices. The equipment and materials not always meet requirements imposed to them for quality and reliability that is shown further in the form of the latent defects at operation. In this regard the scientific and technical research problem consists in development of new theoretical and methodological approaches, technical means of information monitoring and early diagnostics of risks of fire-dangerous situations in power supply networks of technically difficult objects providing anticipatory adoption of reasonable and timely decisions on identification of risk of a fire-dangerous situation in electric networks of technically difficult objects. One of solutions of an objective is use of mnogoagentny approach. In work the electric network is understood as the mnogoagentny distributed information and measuring system which functioning is described through the distributed and concentrated physical and information parameters attached to existential distribution of consumers of loading. It is shown that territorial distribution of sensors and sensors, heterogeneity, multiconnectivity and dynamism of arising fire-dangerous situations in electric networks and protective measures cause considerable complexity of algorithms of the solution of a task of the analysis of all set of controlled parameters of electric networks. Novelty of the developed method of an assessment of risk of emergence of fire-dangerous situations consists in application of hybrid sets of the experimental data obtained as a result of aggregation of vectors of observed and not observed parameters at the level of agents in a mnogoagentny network. Novelty of model of the intellectual agent for an assessment of risk of a fire-dangerous situation consists in a combination of operation of blocks of the analysis of retrospective data, processings of the knowledge realizing algebraic and structural transformations of hybrid data sets for an assessment of individual fire risk. Results of experiment show that efficiency increase due to application of the developed method and the monitoring device "Control-1" leads reaction of system of fire safety on the average to increase in probability of evacuation of people with 0,841 to 0,999 that in turn results in standard values of individual fire risk for considered rooms 106 year1. Thus, prospects of the direction stated during researches, and separate results, such as early diagnostics supernumerary fire-dangerous situations are confirmed.